Performance review conversations that actually land

Most performance reviews are theatre. Both people know the rating before they sit down, both know the conversation will not change it, and the meeting exists so the form can be signed. That version is survivable, and it wastes the one scheduled hour a year in which a person is fully prepared to talk about their own work.

What makes a review fail before it starts?

Surprises.

If the review contains information the person has not heard from you already, the meeting will be about the fact that you did not tell them, and rightly. Nothing else will land. They will spend the following week rerunning conversations in which you said things were fine.

The fix is not in the review. It is in the eleven months before it, and specifically in having the small conversations at the time. Where something has gone wrong during the year, the structured version is the underperformance conversation, and having had it means the review is a summary rather than a revelation.

If you are reading this a week before a review and you know it contains a surprise, the least bad option is to say so immediately: "There's something in here I should have raised in June and didn't. I want to apologise for that before we start."

When should you say the rating?

In the first two minutes.

Managers hold the rating back, usually meaning to build a case for it first. What actually happens is that the person spends the whole build-up trying to work out where it is heading, hears none of the evidence, and interprets every neutral sentence as a clue. By the time the number arrives they have already imagined something worse or better.

Say it, then explain:

"The rating is 'meets expectations'. I want to walk through why, and then spend most of our time on what would move it. Ask me anything as we go."

The person can now actually listen, because the thing they were braced for has already happened.

What makes the evidence usable?

Specific, dated, and drawn from the whole year rather than the last six weeks.

Recency is the most common distortion in reviews. The autumn project is vivid and the spring one has faded, so the rating quietly becomes a judgement about the last quarter. The defence is to keep a running note through the year — four lines a month per person is enough — and to reread it before writing anything.

Each point should name the thing, the date, and the effect. Not "strong communicator" but "took over the Henderson relationship in March after it had gone quiet, and it's now the account with the most consistent forecast." The first is a compliment. The second is a piece of evidence the person can recognise, remember, and repeat.

The same rule applies to criticism, and it is where reviews most often go soft. "Could be more proactive" is unusable. "Twice this year — the June migration and the September launch — a risk you'd spotted didn't reach me until it had already landed" is something a person can do something about.

How do you handle disagreement about the rating?

Settle the facts first, and be willing to lose on some of them.

Someone who disputes their rating is usually disputing one or two specific judgements underneath it rather than the number itself. Ask which: "Which part of this doesn't match how you saw the year?" Then deal with that piece on its own evidence.

If they are right, say so and correct the written record. A manager who never revises anything in a review meeting has made the meeting pointless, and everyone can tell.

If they are wrong, say that too, once, and explain what would have made the difference. Repeating it three times converts disagreement into resentment and buys nothing. Where the reaction turns from disagreement into defensiveness, the specific moves are in what to do when someone gets defensive.

What if they expected a promotion?

Deal with it directly rather than letting it sit under the conversation.

A person who thought this review was going to be the one that moved them up is not listening to your development feedback; they are waiting for the part that explains why it did not happen. Name it early: "I know you were hoping this would be the year. It isn't, and I want to be specific about why."

That conversation has its own shape, and it is worth running properly rather than squeezing into the last five minutes of a review — how to tell someone they are not getting promoted sets out the sequence.

What if you are reviewing someone who is better than you at the work?

Review the outcomes and the behaviours, and be straightforward about the limits of your judgement on craft.

New managers frequently inherit someone more technically capable than they are, and handle it in one of two bad ways: avoiding any judgement of the work, which makes the review empty, or asserting one they cannot support, which destroys credibility in a sentence.

The honest version: "I'm not going to pretend I can assess the quality of your architecture decisions the way a peer could. What I can tell you is what I see — the migration landed on time, two people said your review comments were the most useful they got this year, and the one thing I'd change is how late the risks reached me."

Then use the people who can judge the craft. Ask two colleagues for specific input before you write anything, and say in the meeting that you did. Nobody expects their manager to be the best engineer in the room. They do expect the assessment to be grounded in something.

How much time goes on next year?

About half.

The past section of a review is essentially fixed by the time you are in the room. The forward half is the only part where the meeting can change anything, and it is routinely the part that gets ten rushed minutes at the end because the retrospective overran.

Aim to leave with two or three things, not eight. For each: what it is, what good would look like, and when you will look at it. Then put the first check-in in the calendar before the meeting ends, because a development plan without a date is a wish.

Should the review contain praise?

Yes, and it should be as specific as the criticism.

General praise — "you've had a great year", "you're a real asset" — is pleasant and evaporates. Specific praise is remembered for years, because it tells someone which of the things they did actually registered. That is genuinely useful information: most people cannot tell which parts of their work are visible.

What praise should not be is structural padding around a criticism. The praise-criticism-praise sandwich is the most reliable way to have someone leave a review believing it went well when it did not.

How do you know the review worked?

Ask them, a week later, what the three things were.

If they can tell you — the rating, the main strength, the main thing to change — the meeting did its job, whatever it felt like. If they cannot, or they name three different things from the ones you intended, that is worth knowing and it is not their failure.

The pattern that makes reviews work is the same one that makes every other management conversation work: say the difficult part early, be specific, and leave with something written down. That structure is set out in full in the complete guide to difficult conversations for new managers.